222Rnand 226Ra:indicators of sea-ice effects on air-sea
gas exchange
KENT A. FANNING and LINDA M. TORRES
Fanning. K. A . & Torres. L. M. 1991: I2'Rn and ""Ra: indicators of sea-ice effects on air-sea gas exchange.
Pp. 51-58 in Sakshaug. E . , Hopkins. C. C. E . & Oritsland. N. A . (eds.): Proceedings of the Pro Mare
Symposium on Polar Marine Ecology, Trondhcim. 12-16 May 1990. Polar Research 10(1).
I2'Rn and 226Radistributions beneath the sea ice of the Barents Sea revealed that ice cover has varied
effects on air-sea gas exchange. Twice, once in late summer and once in late winter, seawater samples from
the top meter below drill holes had zzrRnactivities that were not lower than their '"Ra activities. indicating
the existence of secular equilibrium and a negligible net exchange of lr2Rn and other gases with the
atmosphere. However. seawater in the upper 20-85 m usually exhibited at least somc zzrRn depletion;
22'Rn-to-2'bRa activity ratios tended to have 'ice-free' values (0.3-0.9) in the summer and values between
0.9 and 1.0 in the winter. Integrated '*'Rn depletions and piston velocities in both seasons typically fell in
the lower 25% of the ranges for ice-free seawater. suggesting that a moderate but far from total reduction
in gas exchange is normally caused by ice cover and/or meltwater. The results demonstrate that sea-ice
interference with the oceanic uptake of atmospheric gases such as COz is not well understood and needs
further investigation.
Kent A. Fanning and Linda M. Torres, Department of Marine Science, Uniuersit.v of South Florida, Sf.
Petersburg, Florida 33701, USA
rupted by the preferential escape of radon to the
atmosphere, and the ratio of A222to
is thus
ratios in ice-free oceanic mixed
The global significance of the effects of sea-ice <1.0. A222:A226
cover on air-gas exchange is unresolved at layers are typically 0.5-0.9, averaging 0.76
present. A complete absence of atmospheric-C02 (50.09) in GEOSECS studies (Peng et al. 1979)
penetration through sea ice of the Weddell Sea and 0.75 (50.07) in Transient Tracers' studies
was invoked to help explain why sinking Antarctic (Smethie et al. 1985). The lowest observed ratios
Bottom Water appears to carry away only 15% are -0.3 in the Bering Sea (Glover & Reeburgh
of the anthropogenic excess C 0 2 that enters the 1987). Piston uelocify, which is defined as the
surface ocean (Poisson & Chen 1987). A partial thickness of a sea water column exchanging gas
rather than a complete reduction of gas exchange with the atmosphere per unit time (Broecker &
across sea ice was implied by water-mass studies Peng 1982), ranges over nearly two orders of
that found freons F-11 and F-12 to be 20-37% magnitude in ice-free seawater (Table 1).
The work reported here was a reconnaisance
undersaturated in the surface Arctic Ocean (Krysell & Wallace 1988; Wallace et al. 1987). These of 222Rnand 226Radistributions near sea ice. Our
studies emphasized the consequences of high- objectives were to obtain new insights into the
latitude gas exchange on oceanic solutes but they mechanisms of air-sea gas exchange in ice-bearing
did not determine gas-exchange parameters for waters and to estimate the 222Rndepletions and
piston velocities in the presence of sea ice for
the air-seawater (or air-ice-seawater) interface.
222Rnis a quantitative tracer of air-sea gas comparison to values for the same exchange parexchange on time scales of a few days (Broecker ameters in the absence of sea ice.
& Peng 1971, 1974, 1982; Peng et al. 1979; Smethie et al. 1985; Glover & Reeburgh 1987). In a
parcel of seawater isolated from exchange with
Materials and methods
the atmosphere or underlying sediments, activities of 222Rn(A222)and its parent 226Ra(A?2h) The work was done in April 1986 (late winter),
will become equal and attain a condition called and September 1988 (late summer), aboard the
secular equilibrium. Near the sea surface in the Norwegian Coast Guard icebreaker K/V ANDENES
ocean's mixed layer, secular equilibrium is dis- during the Norwegian Pro Mare program on the
Introduction
52 K . A . Fanning & L . ,M. Torres
Table 1 Comparison ot :Rn gas exchange parameters in the Barents SCJ i n late winter (April) and late summer (September)
w i t h 2z2Rngas exchange parameters elsewhere in the ocean
Elsewhere
Barents Sea
Late winter
hydrostaiions
Depth of !'?Rn depletion ( m )
Total ?'!Rn depletion (dpm m
Value( s )
Average
Piston velocity (m d
Value( s)
Average
43
52
110
Late summer
ice station5
Whole
ocean
(Ref 1 )
Tropical
Atlantic
(Ref 2)
Bcring
Sea
(Ref. 3)
7
3
63
10-20
15-20
48 (226)
-
-
280
610
3Jo-680
3jo-450
-
-
-
190-3610
960
330-2190
1239
-
-
0.6
1.5
1.3-2.6
0 9-1.2
-
-
-
0.5-12.7
28
1.4-6.9
-
3.6
0.2-4.9
2.2
1.2
2.8
2.5-5.1
19-2.4
0.6-18.6
33
1.1-6.0
-
3.1
-
')
-
I)
Piston velocity (20°C)
Value(s)
Average
-
-
-
-
References: ( I ) Peng et al (1979) (2) Smcthic et al (1985). (3) Glover & Reeburgh (1987)
Barents Sea. This scheduling permitted a range
of sea-ice effects to be examined because of the
large seasonal variations in Barents Sea ice cover.
The winter cruise (Fig. 1) investigated the
entire water column beneath or adjacent to closepacked winter ice (>90% surface coverage). Sea-
10"
20"
30"
40'
50"E
1 " " " " ' " " " ' " ' 1
c
X ice Slation a2
?V
'Ice Sialion
*
80"N
a3
X ice Stailon L5
m
t
@ =Winter
Slalions (19061
1
I
F I R I . Map of the Barents Sea showing the station locations
during the late -,inter cruise of April 1986. and the late summcr
cruise of September 1988. Approximate ice limits are also
shosn.
water in natural or ship-created open areas was
sampled with 30-liter Niskin bottles to within 10 m
of the bottom. Hydrographic parameters were
determined with a CTD or discrete sampling.
At an ice station ( H l ) , a helicopter served as
transport for collecting seawater below unbroken
ice more than 50 miles from the vessel. A small
( - 10 cm) hole was drilled into the ice, and vacua
in gas-tight, 20-liter PVC sampling bottles were
used to suck seawater from beneath the ice
through weighted tubing (I.D.: 6-12 mm) (Fig.
2).
The late summer cruise (Fig. 1) consisted
entirely of ice stations at which the 1986 icestation protocol was followed (Fig. 2). Surface ice
coverage within the pack was ~ 7 0 % .
2zzRnactivities were determined aboard the
vessel after using the 20-liter PVC bottles as
z22Rn-extractionchambers (Mathieu et al. 1988)
(Fig. 2 ) . Overall 222Rnextraction and counting
efficiency was 85%. After the extractions, the
seawater samples were slowly passed through columns of Mn-oxide-coated acrylic fiber to extract
22hRa(Reid et al. 1979). Later. the columns were
sealed in gas-washing bottles so that the 226Ra
activities of the original samples could be determined from the z22Rningrown after >lmonth of
storage in accordance with the tests and recommendations of Moore (1981). Relative standard deviations of replicate 226Ra analyses
averaged 6%.
Indicators of sea-ice effects 53
INLET VALVE
SIGHT TUBf
GAS
DISPERSION
FRlT
1
a e
v
A
TO
+
=
?
OUTLET
VALVE
i
INLET
VALVE
HOLE DRILLED
WITH AUGER
-1,
WEIGHTED SAMPLE TUBE
ARROWS
SHOW
PATH OF
HELIUM
TOP
OF ICE
dI1
'BOTTOM
OF ICE
D
C
Fig. 2. Diagram of the deployment and use of 20-liter PVC radon sampling and extraction bottles showing: A) an overall view.
B) a cutaway view. C) a bottle in sampling position for sucking seawater through weighted tubing lowered into drill holes at ice
stations. and D) a radon extraction. Vacua in the bottles were also used to suck seawater from 30-liter Niskin bottles at Pro Marc
hydrostations.
face water. Out of sixteen samplings of the upper
85 m, ten had A222:A226
values higher than the
normal range (0.5-0.9), and six had ratios within
The late winter cruise of 1986
it. Five of those (two each at hydrostations 43 and
Upper portions of the winter z2zRnand 226Ra 52 and one at ice station H1) exceeded 0.8. These
profiles (Fig. 3) approached secular equilibrium results suggest that the extensive ice cover
much more closely than reported for ice-free sur- retarded the degree of degassing. One sample in
Results
n
W
I
In
v
E
20
30
I l l
:v)
1
-
OPEN WATER
NEAR ICE EDGE
1
R a Il ?
I
I
t
>
* <5
a
1 '
SHIP-MADE
OPENING
IN THIN ICE
-
-
-
-
-
-
STATION 52
(%o)
400
300
200
100
STATION 43
SALINITY
0 10 20 30 4 0
I
0
Ra I
10
(g)
10
(g)
(%o)
OPEN WATER
WITHIN ICE,
CENTRALBANK
STATION 37
SALINITY
10 20 30 40
0
ACTIVITY
('300)
ICE STA. H1
SALINITY
10 20 30 40
FIR. 3 Profiles of '"Rn activity and "'Ra activity and salinity for thc Barcnts Sea cruisc in lare winter. April 1986 (hydrostations 43. 52 and 37 and icc station H I ) . Dcpths were
measurcd with rcfercncc to the sea surfacc at hydrostations or to the ice surface at ice rtations. Scawatcr rose to within 5 cm of thc ice surfacc in all drill holes. CTD salinity profiles
are shown as solid lincs without data points: discrete salinity mcawrements are Indicated by solid circles (0).
400
300
200
0
ACTIVITY
3
i2.?
3
?
VI
L
Indicators of sea-ice effects 55
the upper 20 meters (the 1-m sample at ice station
H1) showed clear evidence for secular equilibrium, i.e. no reduction in
below
Another 1-meter sample (at hydrostation 43) had
an Az2? value 4% below its A,,, value. This
sample was close to secular equilibrium, but could
also have experienced a small amount of recent
exchange. To our knowledge, the water column
at hydrostation 43 comes the closest to exhibiting
no **'Rn loss to the atmosphere of all oceanic
water columns studied to date. Its lowest
A222:A226
value was 0.93 at 1 0 m ; all others in
the upper 100 m fell between that value and 1.0.
The strongest winter 222Rndepletions were at
hydrostation 37 which had also apparently undergone a recent water-column turnover. A layer in
the upper 20 m with a 40-70% 222Rnsurplus was
both overlain and underlain by seawater with
-20% z22Rndepletion. Large near-bottom '"Rn
enrichments at hydrostations 43 and 52 and the
small near-bottom enrichment at hydrostation 37
indicated that Barents Sea sediments, as Bering
Sea sediments (Glover & Reeburgh 1987) and
other shelf sediments (Fanning et al. 1982), emit
222Rnand label near-bottom water with an A222
value greater than its A226value. Thus the 222Rn
surplus in the seawater at 7-15111 denotes its
recent near-bottom origin, and the 222Rndepletion between 40 and 78m denotes the recent
arrival of air-exposed surface water. The water
column had little density structure to provide
stability, being isohaline (34.5%0 salinity) and
almost isothermal (-1.7 to -13°C). In addition
the water was much shallower than at the other
hydrostations (135 m vs. 300-400 m), and the sea
surface was ice-free. These three conditions probably led to the overturn and the increased outgassing relative to the other hydrostations.
Depths of the 222Rn-deficient zones beneath
the late-winter ice were taken as the depths at
which secular equilibrium appeared or were estimated by linear interpolation between the greatest depth showing a 222Rndepletion and the next
ICE STA. 2
0
10
20
30
10
20
30
SALINITY
(%o)
ICE STA. 3
10
0
30
20
- *.
%
:
-
i \
-
4)
/!la-226
+
:.
10-
I
I
-id
!I
il
-
SALINITY
:I
- &
0
1
Rn-222
Fig. 4. Profiles of 222Rnactivity and **"Raactivity and salinity
for the Barents Sea cruise in late summer. September 1988 (ice
stations 2 and 3). Depths were measured with reference to the
sea surface at hydrostations or to the ice surface at ice stations.
Seawater rosc to within 5 em of the ice surface in all drill holes.
Discretc salinity measurements are indicated hy solid circles
(0).
i
2o
0
10
30
20
SALINITY
(%o)
56 K . A. Funrirng
c(:
L. M . Torres
depth sampled (Table 1). The zones extended
deeper (60-110 m) than usually observed (Peng
et al. 1979) and reached the depletion depths
found after storms (Fanning et al. 1982). Salinities
at hydrostations 43 and 52 had slight increases
from -34.YTr above 5 0 m to -35.1% below
100 m. However the density increases that might
have resulted were partially offset by temperature
increases from - 1 to - 2°C above 50 m to 1.o"C
below 100rn. resulting in a weak water-column
stability that allowed deeper penetration of gasexchange processes. The largest pycnocline that
was found occurred between 100 and 120m at
hydrostation 43 where u, increased by only 0.12.
By contrast, ut changes over 20-meter-thick
pycnoclines in the Bering Sea were -8 times
larger (fig. 3d in Glover & Reeburgh 1987).
beneath ice station 3 (Fig. 4). showed no
reduction of A,?? below A?,,. Since the water
was obviously meltwater (salinity = 6.1%), a
reasonable explanation is that it was confined to
the upper portion of an ice-walled cavity. Otherwise. the mixing processes that were producing
and distributing strongly '??Rn-deficient waters
throughout the region should have destroyed the
strong halocline between 1 and 2 m along with
the associated secular equilibrium. Apparently.
the combination of solid ice walls and meltwater
was capable of restricting '"Rn loss enough to
maintain secular equilibrium despite the substantial air-sea "'Rn exchange indicated by the
normal-to-extensive '??Rn depletions in the top
meter beneath ice stations 2 and 5.
The late sunirner cricise of I988
Discussion
Profiles for this cruise indicated much greater
individual '"Rn depletions (Fig. 4). Most of the
77---Rn-'2hRa
data pairs from ice stations 2 and 3
had activity ratios i n the normal range: 0.5-0.9.
Two had very low ratios of 0.30 and 0.37 (the 2-m
and 3-m values, respectively. from ice station 2)
Not shown is a 1-meter sample taken at ice station
5 (see Fig. 1) with
= 2 . 8 d p m (lOOOL)-'.
= 7.4 dpm (100 L ) - ' . and A ? ? ? :
= 0.38.
These low values confirm previous Bering Sea
findings that water parcels at high-latitude can
undergo considerable ?:?Rn loss during the summer (Glover & Reeburgh 1987).
The summertime "'Rn losses occurred in icemelt-stabilized seawater. North of 79" N in the
region containing the three ice stations (Fig. 1).
Pro Mare CTD casts detected a surface layer in
which temperatures were usually less than
-1.4"C. and salinities ranged from 33.0%~to less
than 1wCr (Fig. 4 ) . The layer was 2 0 m thick at
most locations. Clearly identifiable thermoclines
and haloclines and strong pycnoclines lay between
the layer and deeper waters having temperatures
between -1 and 0°C and salinities >34Cr. The
greater openness of the summer ice pack (s70Qr
coverage) and steady winds from the northwest
quadrant at 3-10 m s - ' apparently enhanced the
degassing in the open areas and the downward
and horizontal mixing of the ??'Rn-deficient
waters in the meltwater layer
One example of under-ice secular equilibrium
was found during the summer cruise. As at the
wintertime ice station H1 in Fig. 3. 1-meter water
---Rn depletions and piston velocities for both
summer and winter were calculated by integrating
depth profiles (Smethie et al. 1985) for all stations
in Figs. 3 and 4 except hydrostation 37 which, due
to turnover. was obviously not at steady state.
Comparisons were then made to parameters
either published in or calculated from previous
studies (Table 1). For hydrostations 43 and 52,
integrations were made between the surface and
the estimated depths of depletion (see above).
Time and equipment constraints prevented a
more detailed sampling of the water column,
resulting in fairly large uncertainties in the
measured thicknesses of the z'rRn depletion
zones at hydrostations 43 and 52. The consequent
errors in the depletions and piston velocities for
those stations in Table 1 are estimated t o be
up to 18% for hydrostations 43 and 30% for
hydrostation 52. These values fall in the ranges
reported by Smethie et al. (1985).
Because the difficulty of sampling through drill
holes precluded the precise determination of
depletion depths during the summer cruise, the
maximum summertime depletion depth was
assumed to be the maximum thickness of the
meltwater layer (20 m). This assumption was
employed because pycnoclines at the base of the
meltwater were 20-40 times the largest 1986 winter pycnocline and 2-5 times the Bering Sea
pycnoclines found by Glover & Reeburgh (1987).
Thus. parameters for ice stations 2 and 3 in Table
1 are shown with ranges. Low values are based
on the average depletions down to the greatest
-
,,,
Indicators of sea-ice effects
depth sampled, and high values on the assumption
that those depletions persisted to 20 meters.
Barents-Sea lZ2Rn depletions and piston velocities fit within the ranges of those parameters
found elsewhere in ice-free seawater. The overall
range of Barents Sea z22Rn depletions (280680 dpm m-I) falls in the lower 14% of the wholeocean range and the lower 19% of the tropicalAtlantic range. The overall range of Barents Sea
piston velocities normalized to 20°C (1.2-5.1 m
d & ' ) falls in the lower 25% of the whole-ocean
range, but the highest normalized Barents Sea
piston velocity is only 15% below the highest
tropical-Atlantic value. Although these comparisons clearly indicate that sea ice restricts gas
exchange, Barents Sea parameters are far from
the zeros to be expected if sea-ice cover routinely
permitted no gas exchange. as required by the
Antarctic excess-COz model of Poisson & Chen
(1987). Our zz2Rn and 226Raresults are much
more consistent with the weaker restrictions on
gas-exchange implied by the moderate freon
undersaturations in the Arctic Ocean (Krysell &
Wallace 1988; Wallace et al. 1987). Even at icecovered hydrostation 43, the upper third of the
water column had A222:Az26ratios that were
uniformly, albeit just slightly, less than unity.
Interestingly, integrated Barents Sea '"Rn
depletions and piston velocities varied little
between summer and winter, suggesting that the
overall effect of sea ice on air-sea gas exchange is
roughly the same in both seasons (Table 1). In
summer, individual percent depletions were high,
but ice meltwater was present to constrain the
depth of exchange. In winter, low water-column
stability permitted the deeper penetration of
222Rnoutgassing to 60-110 m, but the greater ice
cover appeared to restrict the degrees of depletion. Probably there was a deep convection followed by a horizontal advection of ??'Rn-depleted
water associated with the leads that occupy at
least 1%of a sea-ice region, even in winter (Smith
et al. 1990). The distributions of 222Rndepletions
and enrichments at hydrostation 37 (Fig. 3) suggest that the process in those leads might begin
with the turnover of a low-stability water column.
Conclusion
From measurements using a 222Rntracer, our
conclusion is that strong retardation of air-sea gas
exchange by sea ice occurs mainly in isolated
57
regions on the underside of the ice. The ice structures that produce the isolation are unknown.
Normally, even in winter, breaks or other weaknesses in the ice cover seem to be present in
sufficient abundance that the integrated gas
exchange over the surface water column is slightly
less than observed in ice-free seawater. The implication for the possible role of seawater as a sink
for anthropogenic COz is that sea ice is a 'porous'
barrier to the uptake of C 0 2 by high-latitude
surface waters having a p C 0 2 below the
atmospheric value. Poisson & Chen's (1987)
assumption of no gas exchange across Weddell
Sea ice during the formation of Antarctic Bottom
Water is thus open to question, although the lack
of an appreciable p C 0 2 gradient across Weddell
Sea ice may mean that an impermeable ice barrier
is not required to explain the low amounts of
anthropogenic COz in Antarctic Bottom Water.
Until direct studies of '"Rn beneath Weddell
Sea ice are performed in winter, the uncertainty
regarding the influence of Antarctic sea ice on the
fate of atmospheric COz will remain.
Ackriowledgemenrs. - Pro Mare provided logistical support.
Financial support was provided by the Univcrsity of South
Florida Faculty Rcsearch and Creative Scholarship Fund. the
American-Scandinavian Foundation. the U . S . National Science
Foundation (via grant INT 8900496 to Scripps Institution of
Oceanography and grant OCE WI33Y2 to the University of
South Florida), and Pro Mare. C. Rooth supplied valuable
comments and suggestions.
References
Broecker. W. S . & Peng. Ts.-H. 1Y71: The vertical distribution
of radon in the BOMEX area. Earth Planer. Sci. Len. 11,
99-108.
Broecker, W. S . & Pcng. Ts.-H. 1974: Gas exchange rates
between air and sea. Tellur 26. 21-35.
Broecker, W. S. & Peng. Ts.-H. 1982: Tracers in the Sea.
Eldigio Press, Palisades. N.Y
Fanning, K . A,, Brcland. J . A , . 11. & Byrne, R. H. 1Y82:
Radium-226 and radon-222 in the coastal waters of west
Florida: concentration and atmospheric degassing. Science
215. 667-670.
Glover. D. M. & Reeburgh. W . S . 1987: Radon-222 and radium226 in southeastern Bering Sea shelf waters and bediment.
Conr. Sherf Res. 7. 433-456.
Krysell. M. & Wallace. D. W . R 1988: Arcticocean ventilation
studies with a suite of anthropogcnic halocarhon tracers.
Science, 242. 746749.
Mathieu. G. G.. Biscaye. P. E . . Lupton. R. A. & Hammond.
D. E. IY8X: Measurement of radon-222 in natural watcr
systems. Healrli Pltys 55. 989-Y92.
Moore. W. S . IYXI: Radium isotopes in the Chesapeake Bay.
Esruar. Cousr. Shelf Scr. 12. 713-723.
Peng, T5.-H., Broecker. W . S . . Mathicu. G. G. & Li. Y.-€1.
58 K. A. Fanning & L. M . Torres
1979: Radon evasion rates In the Atlantic and Pacific ocean5
as determined during the Gcosecs program. J . Geophvs. Res.
84. 2471-2486
Poisson. A . & Chcn. C:T. 1987. Why is there littlc anthropogcnic CO: in the Antarctic Bottom Watcr? Deep-Sea Res
34. 1255-1275.
Reid. D. F.. Key. R . M. & Schink. D. R. 1979: Radium.
thorium and actinium cxtraction from seawater using an
improved manganese-oxide coated fiber. Eorrh Planer. Scr.
Lerr. 43. 22S226
Smethie. W. M . .J r . . Takahashi. T. & Chipman. D. W. 1985:
Gas exchange and CO: flux in the tropical Atlantic ocean
dctcrmined from ?:'Rn and p C O ? measurements. J. Ceophys.
Res 90.7001-702?.
Smith. S D.. Muench. R. D. & Peasc. C . H . 1990: Polynas and
leads. An ovcrview of physical processcs and environment. J.
Geoplijs. Res. 95. 9461-9379.
Wallace. D. W. R . . Moorc. R. M. & Jones. E. P. iY87:
Ventilation of the Arctic Ocean cold halocline: rates of diapycnal and isopycnal transport. oxygen utilization and primary production inferrcd using ehlorofluoromethane
distributions. Deep-Sea Res. 34. 1957-1979.